IMQ‐induced skin inflammation requires c‐Jun in dendritic cells

To dissect the role of c‐Jun/AP‐1 in skin inflammation, we genetically inactivated c‐Jun using CD11c‐Cre and K5‐Cre‐ER^T2 mice, which target predominantly DCs or keratinocytes, respectively. Additionally, for a broader deletion of c‐Jun, in particular in all hematopoietic cells, we employed the inducible Mx1‐Cre line, where poly I:C treatment induces interferons to activate the Mx‐1 promoter. The treatment scheme for c‐Jun deletion in these mouse models and induction of skin inflammation by Imiquimod (IMQ) is illustrated in (Fig EV1A). Epidermal thickness was used as an initial read‐out for skin inflammation.

Figure EV1

c‐Jun in DCs is required for IMQ‐induced keratinocyte proliferation and differentiation, but is dispensable for skin DC development

AGraphical timeline of IMQ treatment for the different Cre‐Lines. Mice received poly I:C (Mx1‐Cre) or Tamoxifen (K5‐Cre‐ER^T2) for cell‐type specific deletion of c‐Jun and were waxed 2 days before start of IMQ treatment. IMQ was applied daily on the back skin and mice were taken for analysis after 1.5, 3, or 7 days after the first IMQ application, which corresponds to an IMQ treatment of 1, 3 or 7 times, respectively. After 7 days of IMQ application, mice were left untreated for another 3 days to study the resolution of IMQ‐induced skin inflammation. Controls (c‐Jun^{fl /fl}) were treated equally to Mx‐1 or K5‐Cre‐ER^T2 mice in all experiments, which is indicated by a color code (bronze = poly I:C and green = Tamoxifen) here and for all Figures henceforth.

B–DValidation of the deletion efficiency of c‐Jun in the Mx1‐Cre, the CD11c‐Cre and K5‐Cre‐ER^T2 mouse model. Immunofluorescence of c‐Jun (green), K5 (red) and CD45 (white) (B), of c‐Jun (green) and CD11c (red) (C) and of c‐Jun (green) and K5 (red) (D) was performed in back skin from the indicated genotypes. DAPI was used to counterstain. Arrows indicate CD45⁺ (B) or CD11c⁺ skin cells (C). Magnification: 25× (B), 40× (C), 40× (D). Scale Bar: 50 µm.

E, FRatio of K5 to K10 positive areas (E), and number of Ki‐67⁺ cells (F) in epidermis of indicated mice was analyzed by immunofluorescence performed as described in Appendix Fig S1A and B. Three randomly chosen fields per section were analyzed for each sample (n = 6–10 (E), n = 5–12 (F); 2–4 independent experiments)

G–IFlow cytometry of DC subsets in total skin of indicated mice after 2, 3 and 5d of daily IMQ treatment. CD11b⁺ DC cells were defined as MHCII⁺CD172⁺CD11b⁺CD64⁻ (G), CD103⁺ DC as MHCII⁺XCR1⁺CD103⁺CD64⁻ (H) and Langerhans cells as MHCII⁺CD172⁺CD207⁺ CD64⁻ (I) among live, single, CD45⁺ cells. Graphs show DC subsets as % of live, single cells (n = 6–12 (G), n = 4–11 (H), n = 5–13 (I); 2–4 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by one‐way ANOVA with Tukey (E, F) or Bonferroni multiple comparison test (G‐I). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

c‐Jun ^Δ/Δ Mx1‐Cre mice had significantly less IMQ‐induced skin inflammation compared to poly I:C‐treated c‐Jun ^fl/fl controls (Figs 1A and EV1B). This attenuated inflammatory response to IMQ was recapitulated by DC‐specific deletion of c‐Jun (c‐Jun ^Δ/Δ CD11c‐Cre) (Figs 1B and EV1C), while the keratinocyte‐specific c‐Jun deletion (c‐Jun ^Δ/Δ K5‐Cre‐ER^T2 mice) responded to IMQ similar to Tamoxifen treated c‐Jun ^fl/fl controls (Figs 1C and EV1D). Given that c‐Jun seems to be required downstream of TLR7 signaling to mediate the inflammatory response in DCs, we focused our in vivo analyses on c‐Jun ^Δ/Δ CD11c‐Cre mice.

Figure 1

IMQ‐induced skin inflammation requires c‐Jun in dendritic cells

A–CEpidermal thickness of the back skin was analyzed at the indicated time points in c‐Jun ^Δ/Δ Mx1‐Cre (A), c‐Jun ^Δ/Δ CD11c‐Cre (B) and c‐Jun ^Δ/Δ K5‐Cre‐ER^T2 mice (C) on hematoxylin and eosin (H&E) stained sections (n = 5–21 (A), n = 6–29 (B), n = 5–18 (C); ≥ 2 independent experiments).

DH&E stained sections of back skin from indicated mice treated with IMQ for 2, 3, or 5 days. Bright‐field images, Magnification: 20×, Scale bar: 100 µm.

ELayers of epidermis (left) and Trans‐epidermal water loss (TEWL) (right) were analyzed in the back skin of indicated mice (Left: n = 8–14, right: n = 6–14; ≥ 2 independent experiments).

F–HFlow cytometry of total back skin after 2, 3, and 5 days of IMQ treatment. Analyzed were dermal γδ T cells (γδ TCR^int+) (F), monocytes (CD11b⁺Ly6C^hi) (G), and neutrophils (CD11b⁺Ly6G⁺) (H). Graphs show immune cells as % of live, single cells (n = 10–28 (F), n = 10–21 (G), n = 4–25 (H); ≥ 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by one‐way ANOVA with Tukey (A‐C, and E) or Bonferroni multiple comparison test (F‐ H). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

To better understand the role of c‐Jun in the onset of skin inflammation, we analyzed the hallmarks of IMQ‐induced skin inflammation, which are acanthosis (epidermal thickening), loss of barrier integrity, and infiltration of immune cells such as neutrophils and dermal γδ T cells, the major IL‐17‐producing immune cell in murine skin. c‐Jun ^Δ/Δ CD11c‐Cre mice showed a marked reduction in acanthosis and were protected from IMQ‐induced epidermal barrier breakdown as measured by trans‐epithelial water loss (TEWL) (Fig 1D and E). In addition, we observed reduced keratinocyte proliferation (less Ki67⁺ nuclei, less proliferative K5⁺ to differentiated K10⁺ cells) in c‐Jun ^Δ/Δ CD11c‐Cre mice compared to control mice (Fig EV1E and F and Appendix Fig S1A and B). Infiltration of dermal γδ T cells (γδ TCR^int+), monocytes (CD11b⁺Ly6C^hi), and neutrophils (CD11b⁺Ly6G⁺) following IMQ treatment was also significantly decreased in c‐Jun ^Δ/Δ CD11c‐Cre mice (Fig 1F–H), while levels of the two major dermal DC subsets, CD11b⁺ DCs and CD103⁺ DCs, and of epidermal Langerhans cells remained unchanged (Fig EV1, EV2, EV3, EV4, EV5). These results suggest that c‐Jun does not directly control skin DC development, but rather their effector function downstream of TLR7 signaling, which is required for IMQ‐induced skin inflammation.

Figure EV2

Control of cytokine expression in IMQ‐treated skin and BMDCs by c‐Jun

Immunofluorescence of c‐Jun (green), CD11c (red) and DAPI in mouse back skin of indicated mice treated with IMQ for 8 h. Insets give an enlarged view of the framed area. Scale bar: 10 µm, Magnification: 40×.

qRT–PCR detection of chemokines Cxcl1, Ccl20, antimicrobial proteins S100a8, S100a9 and the pro‐inflammatory cytokines Tnfa, Il22, Il6 and Il1b in total RNA isolated from the back skin of indicated mice treated with IMQ for 32 or 48 h (n = 9–20; 3–6 experiments).

qRT–PCR mRNA expression analysis of indicated cytokines in BMDCs. Bar graph shows fold change of targets in IMQ (4 h) stimulated BMDCs relative to LAL. Asterisk shows significance between the genotypes in IMQ‐treated samples (n = 3–7; 2–3 independent experiments).

Immunofluorescence of CD11c (green), IL‐23p19 (red) and DAPI in indicated BMDCs stimulated with IMQ for 5 h in the presence of Brefeldin A. Scale bar: 25 µm, Magnification: 60×.

Gating strategy to sort DCs (CD45⁺CD11c⁺ MHCII⁺), T cells (CD3ε⁺), granulocytes (Gr‐1⁺) and non‐immune cells (CD45⁻) from back skin treated with IMQ for 8 h.

qRT–PCR detection of Il23p19 mRNA expression levels in CD3ε ⁺, Gr‐1⁺ and CD45⁻ cells sorted from back skin as described in (E) (n = 4–6, 3 independent experiments).

IL‐23p19 and IL‐12p40 expression in BMDCs was analyzed by intracellular flow cytometry. BMDCs were stimulated with IMQ for 4 h in the presence of Brefeldin A. Representative plots shown are pregated on live, single, CD45⁺ and CD11c⁺ cells.

Bar graph shows % of live, single, CD45⁺, CD11c⁺ cells (% DCs) expressing IL12‐p40 (n = 7–10; 3 independent experiments).

qRT–PCR detection of Il23p19 mRNA expression in BMDCs pretreated with MG‐132 (NF‐κBi, 20 µM) for 1 h and stimulated with IMQ for 4 h (n = 5–6; 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by unpaired, two‐tailed t‐test (B, C, and F) or one‐way ANOVA with Tukey multiple comparison test (H, I). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Figure EV3

Validation of the therapeutic potential of JNK inhibition in psoriasis‐like mouse models

A–CBack skin of c‐Jun ^fl/fl, c‐Jun ^Δ/Δ CD11c‐Cre, Tlr7 ^−/− and Tlr7 ^−/− c‐Jun ^Δ/Δ CD11c‐Cre mice was treated with IMQ and JNKi or vehicle for 5 days. Shown are representative H&E stained sections (Bright‐field images, Magnification: 20×, Scale bar: 100 µm) (A), immunofluorescence staining’s of Ki‐67 (green) and DAPI (Scale bar: 100 µm, Magnification: 20×) (B) and of K5 (red), K10 (green) and DAPI (Scale bar: 100 µm, Magnification: 20×) (C) in murine back skin. For quantification 2–3 randomly chosen fields per section were counted.

D–FFlow cytometric analysis of cell suspension from back skin of indicated mice and treatment. t‐SNE‐ plots (D) show the cutaneous immune cell phenotype. Populations defined by t‐SNE algorithm were confirmed by conventional gating as described in Appendix Fig S7A and representative plots are shown in (E) for Monocytes and Neutrophils and in (F) for γδ T cells and dendritic epidermal T cells (DETC). Numbers adjacent to marked areas indicate percentage of cell population among CD45⁺ cells (D) and live, single cells (E, F). Representative plots shown are pregated on live, single, CD45⁺ (D) and CD11b⁺ (E) or CD11b⁻ cells (F).

GExperimental design for blocking IL‐23 signaling in the IMQ‐induced skin inflammation model. Back skin of wild‐type mice was treated daily with IMQ and anti‐IL23R antibody (15 mg/kg, i.p.) or isotype control for 5 consecutive days. To compare JNK Inhibitor (15 mg/kg, i.p.) was given (n = 3)

HTrans‐epidermal water loss (TEWL) was analyzed in the back skin at the end of treatment.

IFlow cytometry of total back skin after 5 days of IMQ treatment. Analyzed were dermal γδ T cells (γδ TCR^int+), Monocytes (CD11b⁺Ly6C^hi), and Neutrophils (CD11b⁺Ly6G⁺). Graphs show immune cells as % of live, single cells.

Data information: Data are shown as mean ± SEM. P‐values were calculated by Tukey multiple comparison test (H, I). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Figure EV4

Characterization of c‐Jun co‐expression with CCL2 and IL‐23 in DCs of lesional psoriatic skin

AImmunofluorescence of c‐Jun (green), CD11c (red) and DAPI in psoriatic lesions from two patients and in non‐lesional skin from one patient. Insets give an enlarged view of the framed area. Arrows indicate CD11c⁺ skin cells. Scale bar: 10 µm, Magnification: 40x.

BLinear regression analysis between c‐Jun and Ccl2 in healthy (n = 38), non‐lesional (n = 27) and lesional (n = 28) skin. Expression values were obtained from the GEO Data Set [GSE121212]. A Pearson’s correlation test was performed and r ² and P values are indicated in the plot.

CExpression values of Tlr7, Tlr8, Jnk1, Jnk2, Jnk3, c‐Jun, Ccl2 and Il23p19 in healthy (n = 38), non‐lesional (n = 27) and lesional (n = 28) skin. Expression values were obtained from the GEO Data Set [GSE121212]. Box and whiskers plot: Central band shows median, box extends from the 25^th to 75^th percentiles and whiskers go down to the smallest (min) and up to the largest (max) value.

DRepresentative immunofluorescence of CCL2 or IL‐23p19 (green) and DAPI in healthy and lesional skin. Arrows indicate IL‐23p19⁺ (upper panels) or CCL2⁺ dermal cells (lower panels). Magnification: 40x. Scale bar: 50 µm (n = 2 patient samples).

E, FRepresentative immunofluorescence of CCL2 (E) or IL‐23p19 (F) (white), c‐Jun (green) and the DC markers CD1c or CD14 (red) was performed on lesional psoriatic skin. DAPI was used to counterstain. Arrows indicate triple‐positive cells and an asterisk highlights a representative one that is shown enlarged in a white‐framed inlet. Magnification: 40×. Scale bar: 50 µm (n = 2 patient samples).

Data information: Data are shown as mean ± SEM. P‐values were calculated by Pearson´s correlation test (B) and one‐way ANOVA with Tukey multiple comparison test (C). Statistical significance: ns > 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Figure EV5

Role of JNK‐AP‐1 signaling in the maturation of human mo‐DCs stimulated with R848 or LL‐37‐RNA‐40

A–DHuman mo‐DCs were pretreated with DMSO (1:1,000) or JNK Inhibitor (SP600125, 25 µM) (A, B) or AP‐1 Inhibitor (T‐5224, 20 µM) (C, D) for 1 h and stimulated with R848 for 24 h. Gated CD1a⁺ cells were analyzed for surface expression of CD80 and CD86 by flow cytometry. Shown are representative histograms (A, C) and mean fluorescence intensity (MFI) (B, D) (n = 4 (B); representative result from 2 independent experiments is shown, n = 7 (D); 2 independent experiments).

E–GHuman mo‐DCs were stimulated with LL‐37‐RNA‐40 complex or LL‐37 or RNA‐40 for 24 h. Representative flow cytometry plots and histograms are shown in (E) and mean fluorescence intensity (MFI) of CD80 and C86 is shown in (F). Expression of CCL2 and IL‐23 was analyzed by ELISA (G) (n = 7, 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by paired, two‐tailed t‐test (B, D) or one‐way ANOVA with Tukey multiple comparison test (F, G). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

CCL2‐mediated recruitment of pDCs to IMQ‐inflamed skin depends on c‐Jun in DCs

We next investigated which specific DC subset is affected by the absence of c‐Jun. pDCs express TLR7 and have an essential role in psoriasis pathogenesis (Nestle et al, 2005). They are absent from healthy skin (Gilliet et al, 2004), but are recruited to inflamed skin by a mechanism involving the chemokine CCL2 (Drobits et al, 2012). Thus, we next analyzed specifically pDC recruitment to IMQ‐inflamed skin of c‐Jun ^Δ/Δ CD11c‐Cre mice. The number of pDCs was significantly reduced in c‐Jun ^Δ/Δ CD11c‐Cre mice after IMQ treatment (Fig 2A–C), accompanied by a reduction in CCL2 skin protein expression (Fig 2D). DCs produce CCL2 upon TLR ligation (Mitchell & Olive, 2010), and c‐Jun is a critical regulator of CCL2 expression in cultured fibroblasts (Wolter et al, 2008). We therefore examined whether c‐Jun regulates CCL2 expression also in DCs downstream of TLR7 activation by analyzing TLR7 stimulated bone marrow‐derived dendritic cells (BMDCs), and DCs sorted from IMQ‐treated back skin. We used bone marrow (BM) from c‐Jun ^Δ/Δ Mx1‐Cre mice to generate c‐Jun deficient BMDCs in all experiments, because of the superior deletion efficiency in the bone marrow precursors compared to CD11c‐Cre. As shown in Fig 2 E and F, CCL2 expression in both of these DC populations requires c‐Jun and is controlled by the TLR7/JNK signaling pathway (Fig 2G–I). To demonstrate the importance of CCL2 for pDC migration into the skin, we injected recombinant CCL2 (rCCL2) into mouse back skin. Injection of rCCL2 induced migration of pDCs into the skin and completely normalized skin pDC recruitment in IMQ‐treated c‐Jun ^Δ/Δ CD11c‐Cre mice (Fig 2J). Collectively, these data demonstrate the importance of c‐Jun expression in DCs for the recruitment of pDCs to IMQ‐inflamed skin through expression of the c‐Jun/AP‐1 target gene CCL2.

Figure 2

CCL2‐mediated recruitment of pDCs to IMQ‐inflamed skin depends on c‐Jun in DCs

ASkin infiltration of pDCs (BST‐2⁺ B220⁺ CD11c^int CD11b⁻) was analyzed by flow cytometry. Graph shows immune cells as % of live, single cells (n = 2–15; 1–3 independent experiments).

BRepresentative flow cytometry plots of skin pDCs 12 h after IMQ treatment.

CTotal cell number of skin pDCs defined as in (B) is shown (n = 5–6: 2 independent experiments)

DELISA for CCL2 performed on total back skin treated with IMQ (n = 5–14; 2 independent experiments).

EBMDCs generated from c‐Jun ^fl/fl or c‐Jun ^Δ/Δ Mx1‐Cre BM were stimulated with IMQ for 12 h. CCL2 protein was quantified by ELISA (n = 7–10; 3 independent experiments).

FqRT–PCR detection of Ccl2 mRNA in CD11c⁺MHCII⁺ cells sorted from IMQ‐treated back skin (8 h) (n = 5–12; 3 experiments).

GBMDCs were generated from Tlr7 ^−/−, c‐Jun ^fl/fl, or c‐Jun ^Δ/Δ Mx1‐Cre BM and stimulated with IMQ for 4 h and/or pretreated with SP600125 (JNKi, 25 µM) for 1 h. Ccl2 mRNA was quantified by qRT–PCR (n = 4–18; 2–4 experiments).

H, IWestern blot analysis of Phospho‐ or total JNK and c‐Jun in BMDCs of indicated genotype pretreated with DMSO (1: 1,000) or SP600125 (JNKi, 25 µM) for 1 h and stimulated with IMQ (5 µg/ml) for 15 min.

JFlow cytometry of back skin 12 h after intradermal (i.d.) injection of 1 µg rCCL2 and/or IMQ treatment. Analyzed were pDCs (BST‐2⁺B220⁺CD11c^intCD11b⁻) shown as % of live, single cells (n = 6; 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by multiple t‐test with the Holm–Šídák method (C) or one‐way ANOVA with Bonferroni (A), Kruskal–Wallis (E) or Tukey (D, F, G, J) multiple comparison test. Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

c‐Jun directly regulates IL‐23 expression in DCs

While CCL2 was critical for pDC recruitment following IMQ treatment, it was, in contrast to deletion of c‐Jun in DCs, dispensable for the IMQ‐induced skin inflammation (Appendix Fig S2A–E), suggesting that other factors regulated by c‐Jun are involved. Thus, we next examined the additional functions of c‐Jun in DCs besides the control of CCL2 expression. In DCs sorted from IMQ‐treated back skin, c‐Jun mRNA was the most strongly up‐regulated among the AP‐1 proteins (Fig 3A). Moreover, c‐Jun protein was highly expressed in skin dendritic cells of IMQ‐treated c‐Jun ^fl/fl mice, whereas it was absent in c‐Jun ^Δ/Δ CD11c‐Cre mice (Figs 3B and EV2A). Consistently, c‐Jun in DCs was necessary for the induction of various pro‐inflammatory cytokines and key mediators of IMQ‐induced skin inflammation at specific time points. While only Il23p19 and Il17a were significantly reduced 32h after IMQ treatment in c‐Jun ^Δ/Δ CD11c‐Cre skin (Fig 3C), after 48h there was additionally a marked reduction in mRNA expression of a range of cytokines and chemokines commonly up‐regulated after IMQ application in the skin (van der Fits et al, 2009; Van Belle et al, 2012), including Il22, the chemokines for lymphocytes (Ccl20) and neutrophils (Cxcl1), antimicrobial proteins (S100a8/9), and pro‐inflammatory cytokines, like Tnfa and Il‐6 (Fig EV2B). Accordingly, compared to controls, IL‐23 and IL‐17A levels were not increased by IMQ treatment in the skin of c‐Jun ^Δ/Δ CD11c‐Cre mice (Fig 3D). These results demonstrate that c‐Jun in DCs is necessary for the induction of various pro‐inflammatory cytokines that mediate the onset of IMQ‐induced skin inflammation.

Figure 3

c‐Jun directly regulates IL‐23 expression in DCs

Heat map visualization of AP‐1 family member expression in DCs sorted from back skin of indicated mice after 4 h IMQ stimulation as quantified by qRT–PCR. Color code (red = up, blue = down‐regulation) shows fold change of Ap‐1 mRNA to untreated c‐Jun ^fl/fl DC control. Asterisk shows significant differences in IMQ activated to control c‐Jun ^fl/fl DCs (n = 4–14; 2–4 independent experiments).

Graph shows the percentage of c‐Jun positive skin DCs quantified by immunofluorescence performed as described in Fig EV2 (A) (n = 5, 3 experiments).

qRT–PCR detection of Il23p19 and Il17a in total RNA isolated from the back skin of indicated mice treated with IMQ for 32 or 48 h (n = 10–21; 3–6 experiments).

IL‐23 and IL‐17A protein was quantified by Luminex multiplex assay in cutaneous lysates of indicated mice after IMQ treatment (48 h) (n = 5; 2 independent experiments).

qRT–PCR detection of Il23p19 mRNA expression levels in BMDCs of indicated genotype stimulated with IMQ for 4 h and/or pretreated with JNKi (SP600125, 25 µM, 1 h) (n = 4–14; 2–5 independent experiments).

IL‐23 protein was quantified by ELISA (1 × 10⁶ BMDCs/well) and by intracellular flow cytometry (IL‐23p19⁺p40⁺ cells among live, single, CD45⁺, CD11c⁺ cells) in c‐Jun ^fl/fl and c‐Jun ^Δ/Δ Mx1‐Cre BMDCs stimulated with IMQ for 4 h (Left: n = 10–13; Right: n = 9–10; ≥ 3 independent experiments).

qRT–PCR detection of Il23p19 mRNA expression levels in CD11c⁺MHCII⁺ cells sorted from IMQ (8 h)‐treated back skin of indicated mice (n = 5–12; 3 experiments).

Genomic location and sequence of a putative AP‐1 binding site in the mouse promoter of IL‐23p19. Blockade of c‐Jun/AP‐1 TF binding by inhibition of phosphorylation (JNKi, red), or mutation of the binding sequence (mut 1 = T ‐> C; green | mut 2 = TG ‐> AT; blue) reduces IL‐23p19 promoter activity (see Fig 4J).

c‐Jun chromatin immunoprecipitation of c‐Jun ^fl/fl and c‐Jun ^Δ/Δ Mx1‐Cre BMDCs stimulated with IMQ for 2 h (n = 4–6; 3 independent experiments).

Luciferase activity was quantified in lysates of RAW 264.7 cells transfected with pGL3 basic vector harboring wild‐type (wt) or mutated (mut1 or 2) IL‐23p19 promoter. Transfected cells were pretreated with JNKi (SP600125, 25 µM, 1 h) and stimulated with IMQ (6 h). Results are shown as fold change to pGL‐3‐wt transfected, LAL (Limulus amebocyte lysate)‐treated RAW264.7 cells (n = 4; 2 independent experiments).

IL‐17A production by dermal γδ T cells was analyzed in IMQ‐treated skin (Day 2) of indicated mice by intracellular flow cytometry. Shown are representative plots (left) and a bar graph (right) of IL‐17A⁺ cells pregated on live, single, CD45⁺, γδ TCR^int+ cells (n = 7–10; 3 independent experiments).

Epidermal thickness of back skin was quantified in H&E stained sections of indicated mice treated with rIL‐23 (i.d., 1 µg) and/or IMQ for two consecutive days (n = 4–7; 3 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by unpaired, two‐tailed t‐test (A, C) and one‐way ANOVA with Tukey multiple comparison test (B, D‐G, I‐L). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

We next identified direct transcriptional targets of c‐Jun in IMQ‐stimulated BMDCs among the cutaneous, inflammatory mediators that were reduced in the absence of c‐Jun (Figs 3C and EV2B). We found c‐Jun to be a positive regulator for mRNA expression of the cytokines Il23p19, Il22, Il6, and the chemokines Ccl20 and Cxcl1 in BMDCs (Fig EV2C). Since IL‐23 is a major driver of IMQ‐induced skin inflammation and is predominantly produced by CD11c⁺ DCs (Wohn et al, 2013; Riol‐Blanco et al, 2014), we considered that IL‐23 may represent the critical, direct transcriptional target of c‐Jun/AP‐1 essential for the onset of IMQ‐induced skin inflammation. Closer investigations revealed that IL‐23 expression and protein levels in BMDCs are regulated by TLR7/JNK/c‐Jun signaling as measured by qRT–PCR (Fig 3E), by ELISA and intracellular flow cytometry (Fig 3F), and by immune‐fluorescence (Fig EV2D). In addition, AP‐1 inhibition with the small molecule, T‐5224, completely abolished IL‐23 and reduced CCL2 expression in IMQ‐stimulated BMDCs (Appendix Fig S3A–D). Deletion of c‐Jun also resulted in a significant reduction of Il23p19 mRNA expression in DCs sorted from IMQ‐treated back skin (Fig 3G). Furthermore, we sorted non‐immune (CD45⁻), myeloid (Gr‐1⁺), and T (CD‐3ε⁺) cells and confirmed DCs as the most important source of Il23p19 mRNA after IMQ application (Fig EV2E and F). In contrast, IL‐12p40, the second subunit of the hetero‐dimeric cytokine IL‐23, remained unchanged in the absence of c‐Jun (Fig EV2G and H). Inhibition of NF‐κB, which is known to regulate IL‐23 expression in DCs (Carmody et al, 2007; Zhu et al, 2017), led to a reduction of Il23p19 mRNA similar to c‐Jun deletion (~50%) (Fig EV2I), suggesting that both transcription factors contribute to IL‐23 regulation.

Analysis of the IL‐23p19 promoter sequence revealed several c‐Jun binding sites and chromatin immune‐precipitation assay demonstrated a significant increase in c‐Jun binding to the proximal promoter of IL‐23p19 in c‐Jun ^fl/fl DCs after IMQ treatment compared to c‐Jun‐deficient DCs (Fig 3H and I). A luciferase reporter assay further confirmed the critical role of c‐Jun/AP‐1 for IL‐23 expression. Blockade of JNK signaling or mutation of the putative AP‐1 binding site significantly reduced IL‐23p19 promoter activity in IMQ‐stimulated, transfected RAW 264.7 cells (Fig 3J).

IL‐23 induces IL‐17A production in γδ‐T cells, which contributes to IMQ‐induced inflammation (Tortola et al, 2012). Consistently, analysis of dermal γδ‐T cells of c‐Jun ^Δ/Δ CD11c‐Cre mice treated with IMQ revealed reduced IL‐17A production compared to control mice (Fig 3K). Next, we injected recombinant IL‐23 (rIL‐23) in the back skin of mice treated with IMQ. Importantly, rIL‐23 restored the response to IMQ in c‐Jun ^Δ/Δ CD11c‐Cre mice (Fig 3L). However, rIL‐23 injection was not sufficient to induce CCL2‐dependent pDC migration to the skin (Appendix Fig S4A). Consistently, rIL‐23 treatment only weakly induced Ccl2 mRNA expression, whereas Il17a mRNA was induced strongly (Appendix Fig S4B). In addition, although BM‐pDCs expressed mRNA for the IL‐12 receptor subunits Il‐12rb1 and Il‐12rb2 and could be stimulated with rIL‐12, they lacked expression of the IL‐23 receptor‐specific subunit Il‐23r, and rIL‐23 failed to induce Gzmb and Trail mRNA expression (Appendix Fig S4C and D). To exclude that the observed effects were occurring only in response to IMQ, we investigated whether c‐Jun is critical downstream of TLR7 signaling after stimulation with the non‐Imidazoquinoline, adenine analog compounds, CL264 and CL307. IL‐23, as well as CCL2 protein expression, were significantly reduced in c‐Jun‐deficient BMDCs stimulated with CL264 and CL307 (Appendix Fig S5A–D). Taken together, these results demonstrate that the TLR7/JNK/c‐Jun signaling axis controls IL‐23 production in DCs, providing a molecular explanation for the attenuated skin inflammation observed in c‐Jun ^Δ/Δ CD11c‐Cre mice.

c‐Jun is essential in conventional type‐2 DCs to control CCL2 and IL‐23 expression

The skin harbors different DC subpopulations, also known as conventional type‐1 and type‐2 DCs with unique functional properties (Haniffa et al, 2015). While cDC1 promote Th₁ immunity, cDC2 drive Th₂ and Th₁₇ responses (Collin & Bigley, 2018). We next asked whether c‐Jun promotes CCL2 and IL‐23 expression in a distinct cutaneous DC subset. To specify the relevant DC subpopulation and to investigate, whether macrophages contribute to the observed phenotypes, we sorted cDC1, cDC2, Langerhans cells, and macrophages from murine wild‐type skin. IMQ‐induced c‐Jun, Ccl2, and Il‐23p19 expression was most prominent in the cDC2 subset, but Ccl2 was also induced, to a lesser extent, in cDC1, LCs, and macrophages. All three genes were poorly induced in macrophages after IMQ treatment (Fig 4A). However, macrophages had the highest basal gene expression of Ccl2 (Appendix Fig S6A and B). When compared to c‐Jun ^fl/fl mice, sorted cDC2 subsets from c‐Jun ^Δ/Δ CD11c‐Cre mice expressed significantly reduced levels of Ccl2 and Il23p19, whereas the expression of these genes was not affected in sorted macrophages (Fig 4B and C). Moreover, Ccl2 and Il23p19 were also expressed at similar levels in CD11c expressing macrophages sorted from c‐Jun ^fl/fl and c‐Jun ^Δ/Δ CD11c‐Cre mice confirming that these macrophage subsets was not contributing to the observed phenotype. In fact, c‐Jun expression was also not affected in CD11c expressing macrophages from c‐Jun ^Δ/Δ CD11c‐Cre mice as shown by qRT–PCR, demonstrating that c‐Jun deletion by CD11c‐Cre had not occurred in these cells (Fig 4D and E). These results demonstrate that the phenotype observed is most likely due to c‐Jun activity in the cDC2 subset.

Figure 4

c‐Jun is essential in conventional type‐2 DCs to control CCL2 and IL‐23 expression

DC subsets (cDC1, cDC2), macrophages (MP), and Langerhans cells (LC) were sorted from IMQ‐treated back skin (5 h). c‐Jun, Ccl2, and Il23p19 mRNA was analyzed by qRT–PCR. Sort strategy is shown in (B) (10 mice were pooled per condition, 1 experiment).

Sort strategy for the separation of MPs (CD45⁺Lin⁻CD64⁺CD172a⁺CD11b⁺), and the DC (CD45⁺Lin⁻CD64⁻CD11c⁺MHCII⁺) subsets cDC1 (XCR1⁺CD11b⁻ DC), cDC2 (XCR1⁻CD11b⁺CD24^−/intCD326⁻) and LC (XCR1⁻CD11b⁺CD24⁺CD326⁺) in IMQ‐treated (5 h) murine back skin.

qRT–PCR detection of c‐Jun, Ccl2, and Il23p19 in MPs and cDC2 sorted as described in (B) from indicated mice (MPs: n = 7–9; 3 independent experiments, cDC2: n = 5–10; 2–3 independent experiments and back skin from 2 mice was pooled).

Sort strategy for CD11c expressing MPs (CD45⁺Lin⁻CD11c^int/lowMHCII⁺CD64⁺CD11b⁺CD172a⁺) is shown. Murine skin was treated for 5 h with IMQ.

Expression of indicated targets was analyzed by qRT–PCR in cells sorted as described in (D) (n = 3; Back skin from 2 mice was pooled).

Data information: Data are shown as mean ± SEM. P‐values were calculated by unpaired, two‐tailed t‐test (C, E). Statistical significance: ns > 0.05, *P < 0.05. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Blocking JNK/c‐Jun signaling ameliorates IMQ‐induced skin inflammation

Next, we explored the therapeutic potential of blocking the TLR7/JNK/c‐Jun signaling cascade in the IMQ‐induced psoriasis‐like skin inflammation model. c‐Jun^{fl /fl} mice were treated daily with IMQ and the JNK inhibitor (JNKi), SP600125. Tlr7 ^−/−, c‐Jun ^Δ/Δ CD11c‐Cre, and Tlr7 ^−/− ‐ c‐Jun ^Δ/Δ CD11c‐Cre mice were used as controls (Fig 5A). JNKi treatment ameliorated IMQ‐induced skin thickening, trans‐epidermal water loss (TEWL), keratinocyte proliferation, and differentiation comparable to deletion of c‐Jun in DCs in the c‐Jun ^Δ/Δ CD11c‐Cre mouse model. In addition, JNKi treatment of Tlr7 ^−/− and c‐Jun ^Δ/Δ CD11c‐Cre mice showed that these effects were restricted to the TLR7/c‐Jun signaling pathway (Figs 5B–E and EV3A–C). t‐distributed stochastic neighbor embedding (t‐SNE)—analyzes of the cutaneous immune cell populations revealed an IMQ‐induced inflammatory skin phenotype in c‐Jun ^fl/fl mice, which was absent in JNKi treated mice (Fig EV3D). Conventional FACS analysis confirmed the immune populations defined by t‐SNE (Appendix Fig S7A and B) and showed a significantly reduced frequency of monocytes, neutrophils, and γδ‐T cells in JNKi treated, as well as, c‐Jun ^Δ/Δ CD11c‐Cre and Tlr7 ^−/− mice (Figs 5F–H and EV3E and F). Lastly, a protein screen showed that JNKi treatment significantly reduced the pro‐inflammatory cytokines IL‐33 and IL‐17A and the chemokines CCL2 and CXCL1 on therapy endpoint (5d IMQ) compared to the vehicle‐treated control (Appendix Fig S7C).

Figure 5

Blocking JNK/c‐Jun signaling ameliorates psoriasis‐like skin inflammation

AExperimental design for blocking JNK/c‐Jun signaling in the IMQ‐induced skin inflammation model. c‐Jun ^fl/fl, c‐Jun ^Δ/Δ CD11c‐Cre, Tlr7 ^−/−, and Tlr7 ^−/− c‐Jun ^Δ/Δ CD11c‐Cre mice were treated with IMQ and JNK inhibitor SP600125 (JNKi, 15 mg/kg, i.p.) or vehicle for 5 consecutive days.

B–EAfter 5 days of treatment TEWL was measured on the back skin (B), epidermal thickness was analyzed on H&E stained sections (C), epidermal Ki‐67⁺ cells (D) and the ratio of K5 to K10 positive areas (E) were analyzed by immunofluorescence as described in Fig EV3 (B, C) (n = 4–15; ≥ 2 independent experiments).

F–HQuantification of immune cell populations as defined in Fig EV3 (E, F). Results are given as percentage of live, single cells (n = 4–12; 2–4 independent experiments).

IExperimental design for blocking JNK/c‐Jun signaling in c‐Jun/JunB ^Δ/Δ K5‐Cre‐ER^T2 mice. Psoriasis‐like disease was induced by injection of Tamoxifen (i.p., 5 days). Treatment with JNK Inhibitor (15 mg/kg, every other day, i.p.) started on day 14 for 2 weeks.

JRepresentative images of ear, tail and paw of c‐Jun/JunB ^Δ/Δ K5‐Cre‐ER^T2 at the start (Day 14) and end (Day 28) of treatment with JNK Inhibitor or vehicle.

KH&E stained sections of mouse ear from indicated mice. Bright‐field images, Scale bar: 250 µm, Magnification: 10×.

LPsoriasis‐like phenotype was scored from 0 to 4 for ear, tail, paw and snout. Shown is the cumulative score at therapy start (Day 14) and end (Day 28). Non‐responders are c‐Jun/JunB ^Δ/Δ K5‐Cre‐ER^T2 mice treated with JNK Inhibitor, which showed disease‐progression (n = 2–9, 2 independent experiments).

MRepresentative t‐SNE‐ plots (n = 2–3) show the cutaneous immune cell phenotype at therapy end (Day 28). Numbers adjacent to gates give cell populations frequency among live, single, CD45⁺ cells. t‐SNE Legend‐Plot labels the populations defined by t‐SNE‐ algorithm.

NDermal γδ T cells (γδ TCR^int+), monocytes (CD11b⁺Ly6C^hi), and neutrophils (CD11b⁺Ly6G⁺) were analyzed by flow cytometry in mouse ear at the end of treatment (Day 28). Graphs show immune cells as % of live, single cells (n = 2–8, 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by paired, two‐tailed t‐test (L) and one‐way ANOVA with Tukey (B‐H) or Bonferroni multiple comparison test (N). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Figure 6

c‐Jun, CCL2, and IL‐23 are co‐expressed in type‐2/inflammatory DCs in psoriatic lesions

AGene expression values of Cd207, Cd1a, Cd1c, Cd14, Xcr1 in healthy (n = 38), non‐lesional (n = 27), and lesional (n = 28) skin. Expression values were obtained from the GEO Data Set [GSE121212]. Box and whiskers plot: Central band shows median, box extends from the 25^th to 75^th percentiles, and whiskers go down to the smallest (min) and up to the largest (max) value.

BImmunofluorescence of c‐Jun (green), CD1a or CD1c or CD14 (red) and DAPI in psoriatic lesions. Shown is a representative image (Magnification: 25×, Scale Bar: 100µm) with an Inset (Magnification: 63×, Scale Bar: 20 µm, Deconvoluted). Arrows indicate c‐Jun⁺ DCs. Asterisk highlights a DC with prominent c‐Jun expression as shown enlarged in 2 white‐framed windows (DC Marker (red) + c‐Jun (green, left) or DAPI (blue, right)).

CQuantification of c‐Jun⁺ DCs (CD1a, CD1c, or CD14) in psoriatic lesions as shown in (B). Four randomly chosen fields on one section were analyzed for each sample (n = 2 patient samples/condition). Box and whiskers plot: Central band shows median, box extends from the 25^th to 75^th percentiles, and whiskers go down to the smallest (min) and up to the largest (max) value.

D, ERepresentative immunofluorescence of c‐Jun (green), CD1a (red), CCL2 (D), or IL‐23p19 (E) (white) and DAPI in psoriatic lesions. Arrows indicate triple‐positive cells and an asterisk highlights a representative one that is shown enlarged in a white‐framed inlet. Magnification: 40x. Scale bar: 50 µm (n = 2 patient samples).

Data information: Data are shown as mean ± SEM. P‐values were calculated by one‐way ANOVA with Tukey multiple comparison test (A, C). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

Figure 7

JNK/AP‐1 Inhibitors repress CCL2 and IL‐23 expression in human mo‐DCs

AWestern blot analyses of Phospho‐ or total JNK and c‐Jun in human mo‐DCs pretreated with DMSO (1: 1,000) or SP600125 (JNKi, 25 µM) for 1 h and stimulated with R848 (10 µg/mL) for 15 min.

B, CHuman mo‐DCs pretreated with JNKi (SP600125, 25 µM) (B) or AP‐1 Inhibitor (T‐5224, 20 µM) (C) for 1 h were stimulated with R848 (10 µg/ml). CCL2 and IL‐23 were analyzed by ELISA (n = 5–11; 2–3 independent experiments).

DRepresentative flow cytometry plots (left) and histograms (right) of human mo‐DCs stimulated with LL‐37‐RNA‐40 for 24 h are shown. Inhibitors (JNKi; 25 µM or AP‐1 Inhibitor; 20 µM) were given 1 h before stimulation. Plots shown are pregated on live, single, CD1a⁺ cells.

E, FHuman mo‐DCs described in (D) were analyzed for mean fluorescence intensity of CD80 or CD86 by flow cytometry (E) and CCL2 or IL‐23 protein by ELISA (F) (n = 7; 2 independent experiments).

Data information: Data are shown as mean ± SEM. P‐values were calculated by paired, two‐tailed t‐test (B, C, E and F). Statistical significance: ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001. See Appendix Table S3 for exact P‐values.

Source data are available online for this figure.

We tested the therapeutic effects of JNKi in a second independent mouse model of psoriasis based on genetic, inducible and epidermal‐specific deletion of c‐Jun and JunB (c‐Jun/JunB ^Δ/Δ K5‐Cre‐ER^T2 mice) (Zenz et al, 2005). Inhibition of JNK (Fig 5I) ameliorated psoriatic disease in 6 out of 8 mice, whereas disease progressed in 5 out of 7 control mice (Fig 5J–L) with reduced infiltration of monocytes and γδ‐T cells (Fig 5M and N). As we demonstrated IL‐23 signaling to be the crucial pathway activated by JNK/c‐Jun, we next compared the efficiency of JNK inhibition to IL‐23R blockade. Treatment with anti‐IL‐23R antibody alleviated IMQ‐induced skin inflammation as effectively as JNKi treatment (Fig EV3, EV4, EV5). These data show that pharmacological inhibition of JNK aimed to disrupt the pathogenic TLR7/JNK/c‐Jun/IL‐23 signaling axis ameliorates skin inflammation in psoriasis‐like mouse models.

c‐Jun, CCL2, and IL‐23 are co‐expressed in type‐2/inflammatory DCs of psoriatic lesions

Next, we investigated the human relevance of our findings by analyzing the expression of c‐Jun in DCs of psoriatic lesions of patients. Immunofluorescence microscopy revealed co‐localization of c‐Jun protein with CD11c⁺ DCs in lesional dermis. In contrast, in non‐lesional skin, expression of c‐Jun in DCs was weak or absent (Fig EV4A).

To strengthen this finding, we next assessed the expression of different DC markers in a published RNA‐Seq Dataset generated from human psoriatic skin tissue (Data ref: Tsoi et al, 2019b). We found markers for cDC2 (Cd1c) and inflammatory DCs (Cd14) to be increased in lesional psoriasis (Fig 6A), as well as an increase in expression for Tlr7/8, Jnk1, c‐Jun, Ccl2, and Il23p19. Consistently, CCL2 and IL‐23p19 protein was increased in lesional compared to healthy skin. Lastly, c‐Jun expression strongly correlated with Ccl2 expression in lesional, but not in non‐lesional or healthy skin (Fig EV4, EV5). Therefore, we analyzed c‐Jun expression in DCs with CD1a, CD1c, or CD14 markers in healthy, non‐lesional, and psoriatic skin. c‐Jun was expressed equally in CD1a, CD1c, and CD14 expressing cells (Fig 6B) and increased in lesional compared to healthy and non‐lesional skin (Fig 6C). Moreover, co‐expression of c‐Jun with CCL2 or IL‐23 could be detected in CD1a⁺ cells (Fig 6D and E), as well as in CD1c⁺ cells and CD14⁺ cells (Fig EV4E and F). These results suggest that c‐Jun expression in type‐2/inflammatory DCs may also promote psoriasis in patients, by controlling the expression of CCL2 and IL‐23.

JNK/AP‐1 Inhibitors repress CCL2 and IL‐23 expression in human mo‐DCs

Next, we generated human monocyte‐derived DCs (mo‐DCs) to study the importance of Jun/JNK signaling for the expression of CCL2 and IL‐23, the two essential cytokines we identified to be regulated by c‐Jun in murine DCs. mo‐DCs stimulated with the TLR7/8 agonist Resiquimod (R848) showed phosphorylation of Jun/JNK, which was prevented by addition of the JNK inhibitor SP600125 (Fig 7A). Inhibition of Jun/JNK signaling with JNKi or T‐5224 (AP‐1 Inhibitor) reduced CCL2 and IL‐23 expression in R848 stimulated mo‐DCs (Fig 7B and C), but DC maturation, as measured by CD80 and CD86 up‐regulation, was unaffected (Fig EV5A–D). Next, we stimulated human mo‐DCs with the psoriasis‐promoting LL‐37/RNA complex, a disease‐relevant TLR7/8 agonist (Ganguly et al, 2009). LL‐37/RNA complex induced maturation (Fig EV5E and F) and expression of CCL2 and IL‐23 in human mo‐DCs (Fig EV5G), which was inhibited by JNK and AP‐1 Inhibitors (Fig 7D–F). These results demonstrate that c‐Jun in DCs contributes to psoriatic inflammation by regulating CCL2 and IL‐23, two cytokines required for pDC recruitment and activation of IL‐17A producing T cells.

Mice

Mice were kept in the animal facility of the Medical University of Vienna in accordance with institutional policies and federal guidelines. c‐Jun ^fl/fl (Behrens et al, 2002) mice were crossed to mice with Cre recombinase under control of the CD11c (Caton et al, 2007), or Mx1 promoter (Kuhn et al, 1995) or to mice expressing an estrogen‐receptor fusion Cre recombinase under control of the K5 promoter (Brocard et al, 1997). Tlr7 ^−/− (Hemmi et al, 2002) and Ccl2 ^−/− (Lu et al, 1998) have been described previously. Mice were kept in a mixed background (129 Sv×× C57BL/6). Mice had unlimited access to standard laboratory chow and water under a light dark cycle of 12 h in a room with a temperature of 22°C. Mx1‐Cre mice were injected with poly I:C (200 µg, i.p., VWR) twice, at an interval of 5 days, to induce deletion and taken earliest 9 days after the last injection. K5‐Cre‐ER^T2 mice were injected with Tamoxifen (Tx, 1 mg, i.p., Sigma‐Aldrich) for 5 consecutive days. c‐Jun ^fl/fl mice were used as controls in experiments with the respective c‐Jun ^Δ/Δ Mx1‐Cre, CD11c‐Cre, or K5‐Cre‐ER^T2 mice. Wild‐type mice were used as controls for Ccl2 ^−/−. Control mice received the same treatment as experimental mice, e.g. poly I:C (Mx1‐Cre) or Tx (K5‐Cre‐ER^T2).

Imiquimod treatment

Female and male mice, 8–12 weeks of age were used for all experiments. Back skin of mice was shaved (Aesculap GT608) and waxed (Veet, Heidelberg, Germany) 48 h before the start of IMQ treatment. Mice were treated daily with a 5% cream formulation of IMQ (Aldara, Meda Pharma) on the back skin for up to 7 days and left untreated after day 7 for 3 additional days to study the resolution of skin inflammation.

Pharmacological inhibition of JNK‐IL23R signaling in vivo

The JNK inhibitor SP600125 (Abcam) was administered intraperitoneally at a dose of 15 mg/kg. As a control mice received the vehicle (5% DMSO, 5% Tween‐80, and 30% PEG‐300). Alternatively, mice received 15 mg/kg of a monoclonal antibody to mouse IL‐23R (i.p., Merck Research Laboratories) or an isotype mouse IgG1 antibody (i.p., Hölzel).

Psoriasis‐like mouse model

The previously described c‐Jun/JunB ^Δ/Δ K5‐Cre‐ER^T2 mice were kept in a mixed background (129 Sv × C57BL/6 background) for a psoriasis‐like phenotype to develop (Zenz et al, 2005) after injection of Tx (1 mg, i.p.) for 5 consecutive days. 14 days after the first Tx injection administration of JNK inhibitor SP600125 (15 mg/kg, i.p., every other day, Abcam) started for 2 weeks. Psoriasis severity of experimental mice was analyzed similar to a previously reported psoriasis severity scoring system (Glitzner et al, 2014). Shortly, ear, tail, paw, and snout were scored from 1 to 4 (healthy to severest).

Intradermal cytokine injection

One µg of recombinant IL‐23 (BioLegend) or CCL2 (BioLegend) was injected daily into a defined, marked area of the back skin. PBS was injected as a control.

Flow cytometry

Single‐cell suspensions were prepared (see Appendix Methods) and subsequently blocked with anti‐CD16/32 antibody (BioLegend). Cell surface markers were stained with fluorescently labeled antibodies for 30 min at 4°C (see Appendix Table S1). After incubation, cells were washed, filtered, and stained with 7‐AAD Viability Staining Solution (BioLegend) according to the manufactures recommendation to exclude dead cells. Cells were recorded on a LSR Fortessa cell analyzer (BD Biosciences) and analyzes was done with FlowJo software (Version 10.6.1, Treestar).

Intracellular cytokine staining

Cells were incubated with Brefeldin A (BioLegend) for 4 h after stimulation. A fixable viability dye (Zombie Aqua, BioLegend) was used to distinguish dead cells. Cell surface markers were stained before cells were fixed and permeabilized with BD Cytofix/Cytoperm (BD Biosciences). Permeabilized cells were incubated with the appropriate intracellular antibody (see Appendix Table S1) or isotype control for 30 min at 4°C.

Cell sorting

Immune cells were sorted from back skin. For sorting, a single‐cell suspension was prepared and cells were stained with the appropriate fluorescently labeled antibodies. Cells were sorted into TRIzol LS Reagent (Thermo Fisher Scientific) with a FACS Aria III. RNA was extracted with the column‐based miRNeasy Micro Kit (Qiagen) according to the manufacturer’s protocol. cDNA synthesis was performed with Superscript IV Reverse Transcriptase (Thermo Fisher Scientific).

TEWL measurement

TEWL of back skin was measured with a Tewameter® TM 300 probe attached to the MDD4‐display device (Courage + Khazaka) according to manufacturer’s recommendations.

RNA extraction

Back skin was stored in RNALater (Applied Biosystems) at −20°C until use. To isolate RNA, the back skin was mechanically disrupted with a Precellys 24 homogenizer (Bertin, ~ 5,000 g, 2 × 30 s). Cell debris was removed by centrifugation. RNA was isolated with TRIzol Reagent by chloroform/ethanol extraction according to the manufacturer´s protocol (Invitrogen). Protoscript II (New England Biolabs) was used for cDNA synthesis. Real‐time quantitative PCR (qRT–PCR) analysis was performed on an ABI7500‐Fast Real Time PCR system (Applied Biosystems) with Power SYBR Green Master Mix (Applied Biosystems). Primer sequences are listed in Appendix Table S2.

Protein quantification

A piece of back skin was snap frozen and stored at −80°C. Samples were homogenized with a Precellys homogenizer (Bertin) in Precellys tubes (Peqlab). Total protein quantification in skin lysates was done by Bradford protein assay according to the manufacturer´s protocol (Bio‐Rad). 40 (ELISA) or 100 µg (Luminex) of total protein diluted in ELISA assay diluent (BioLegend) with a protease inhibitor cocktail (Roche) were used to perform an ELISA for CCL2 (BD Biosciences) or a Multiplex Luminex assay (Thermo Fisher Scientific) according to manufacturer’s instructions. Analysis of the Multiplex Luminex assay was done on a Luminex MAGPIX System using the xPONENT Software.

Cell culture

To generate BM‐derived immune cells from Mx1‐Cre mice in vitro, deletion was first induced by poly I:C injection in vivo (as described above). Controls were treated equally. BM was then isolated from tibia and femur, and red blood cells were lysed. Isolated cells were cultured in 4 ml RPMI‐1640 (Gibco) with 10% FCS (PAA), penicillin, streptomycin, non‐essential amino acids (Sigma‐Aldrich), sodium pyruvate (Sigma‐Aldrich), β‐mercaptoethanol (Gibco), and GM‐CSF (20ng/mL, Peprotech) for 3 days. On day 3, media was changed. On day 6, non‐adherent cells (BMDCs) were harvested and used for experiments (CD11c⁺ cells > 80%). BMDCs were stimulated with IMQ, CL264, or CL307 (InvivoGen). For inhibition BMDCs were pretreated with SP600125 (InvivoGen; 25 µM) or T‐5224 (20 µM, ApexBio) for 1 h before stimulation. Protein levels of CCL2 (BD Biosciences) and IL‐23 (BioLegend) were analyzed by ELISA in BMDC supernatants. BM‐pDCs were generated as previously described (Drobits et al, 2012).

Chromatin immunoprecipitation

The SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology, #9003) was used to perform a chromatin immunoprecipitation. In short, BMDCs were stimulated with IMQ or LAL for 2 h, fixed with 1% (v/v) formaldehyde for 10 min at RT, cells were lysed, digested with micrococcal nuclease, and ultrasonicated using a Bioruptor Plus (diagenode) sonication device. Sonicated samples were incubated with 10 µg anti‐c‐Jun antibody (Cell signaling, 60A) o/n at 4°C, followed by an incubation with 30 µl CHIP grade magnetic beads (2 h, 4°C). DNA was isolated using spin columns after an elution and digestion of samples with Proteinase K (40 µg). Recovered DNA was analyzed by qRT–PCR with primers specific to the presumptive c‐Jun binding site of the IL‐23p19 promoter (see Appendix Table S2).

Luciferase reporter assay

The IL‐23p19 promoter was cloned into the pGL3‐basic vector (Promega). Site‐directed mutagenesis was performed using the Q5 Site‐Directed Mutagenesis Kit according to the manufacture´s protocol (New England Biolabs). RAW 264.7 cells (ATCC) were cultured in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% FCS (PAA) and re‐plated (6 × 10⁵ cells/well) one day before transfection. Transfection was done with the jetPRIME transfection reagent according to the manufacture´s protocol (Polyplus‐transfection). After transfection (4 h), cells were pretreated (1 h) with JNK Inhibitor (SP600125, 25 µM) and stimulated with IMQ (6 h) before luciferase activity was measured using the Luciferase Assay System (Promega). Mutation primers are listed in Appendix Table S2.

Human monocyte‐derived dendritic cells

Buffy coats from healthy donors were purchased from the Medical University of Graz, Department of Transfusion Medicine, Austria. For the isolation of peripheral blood mononuclear cells, heparinized blood was separated by gradient centrifugation with Lymphoprep^TM (Axis Shield). CD14⁺ monocytes were positively selected according to the manufacturer’s instructions (CD14 MicroBeads, Miltenyi Biotec). For moDC generation, CD14⁺ monocytes (1x10⁶ cells/ml) were cultured for 6 days in RPMI‐1640 (Sigma‐Aldrich) supplemented with 10% FBS and recombinant human cytokines: 35 ng/ml IL‐4 and 100 ng/ml GM‐CSF (PeproTech).

LL‐37/RNA‐40 stimulation

Human mo‐DCs were pretreated for 1 h with JNK inhibitor SP600125 (25 µM, InvivoGen) or the AP‐1 inhibitor T‐5224 (20 µM, ApexBio) before a stimulation with 10 µg/ml of Resiquimod (R848, InvivoGen) or RNA‐40 (10 µg/ml; iba‐lifesciences) in complex with LL‐37 (50 µg/ml; InvivoGen) for 4 or 24 h. For complex formation, RNA‐40 was incubated with LL‐37 for 1 h at room temperature. Human CCL2 and IL‐23 (Thermo Fisher Scientific) was analyzed by ELISA.

Human patient material

Paraffin‐embedded (n = 2) or frozen (n = 2) lesional and non‐lesion (n = 2) materials from biopsy samples were available from psoriasis patients. Healthy skin (n = 2) was collected after plastic surgery. All samples were obtained through informed consent by an approved protocol (Ethics approval 27‐071/ Medical University of Graz Institutional Review Board and 1360/2018/Medical University of Vienna). All experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

Immunofluorescence of human skin

For double immunofluorescence (IF) of c‐Jun with CD1a, CD1c or CD14 paraffin sections (5 µm) were deparaffinized in xylene and rehydrated with decreasing concentrations of ethanol according to standard method. Sections were subjected to HIER antigen retrieval in Target Retrieval Solution pH 6.0 (Agilent/Dako, USA) for 10 min. Slides were allowed to cool for 45 min before rinsing in Tris‐Buffered Saline Tween‐20 (0.5%) (TBS‐T). Sections were blocked with 5% donkey serum, 5% BSA in TBS‐T for 1 hour before incubation with primary antibodies (overnight, 4°C). Slides were washed (3 × 10 min TBS‐T), and appropriate secondary AB was applied for 1 hour. DAPI was used to counterstain and slides were mounted with Dako Fluorescence Mounting Medium (Agilent, Inc., Santa Clara, CA).

For double IF of c‐Jun with CD11c skin biopsies were embedded in optimal cutting temperature compound O. C. T.™ (Sakura), cut (5 µm), fixed in paraformaldehyde, blocked, and incubated with primary and secondary AB as described above.

For triple IF of c‐Jun with DC markers (CD1a, CD1c, CD14) and cytokines (CCL2, IL‐23), frozen skin was acetone fixed, washed (3 × PBS), blocked (5% donkey serum in PBS) for 1 h at room temperature, and incubated with the primary antibodies overnight (4°C). Subsequently, slides were washed (3 × PBS) and incubated with secondary antibodies (1 h, room temperature). Slides were counterstained with DAPI. Images were recorded on a LSM700 Confocal Microscope. ZEN 3.0 image acquisition software was used for image processing. On merged images adjustment of individual color channels was done with Adobe Photoshop. Deconvolution was done with Huygens software.

Human RNA expression data

RNA‐Seq data from (Tsoi et al, 2019a) were analyzed using the processed read‐count data supplied by the submitter (Data ref: Tsoi et al, 2019b). Raw read‐counts from healthy (n = 38), non‐lesional (n = 27), and lesional (n = 28) patient‐derived skin transcriptomes were TMM (Trimmed Mean of M‐values) normalized using the edgeR package (Robinson et al, 2010; McCarthy et al, 2012) from the Bioconductor project to compare expression values.

Statistical analysis

GraphPad Prism 8 software was used for statistical analysis. Unpaired and paired two‐tailed Student’s t‐test, one‐way, or two‐way ANOVA with Tukey or Bonferroni multiple comparison test (selected pairs of columns) were used. Welch’s correction was performed on Student’s t‐test, if variance between groups was significantly different. When possible normality of data was assessed with D'Agostino and Pearson test, outliers were identified by Grubbs´ test or ROUT Method. P values of < 0.05 were considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Exact P values are listed in Appendix Table S3.